[r-t] Big searches

[Mark Davies]

Breaking news - my Big Search has just ticked over the 200 trillion node
mark, at about 860 hours running time, and 67% complete. Could still reach
300 trillion.

It's struck me that the quadrillion-node watermark is within reach with
modern computers and a high-rate, unpruned search. I didn't reserve space on
screen for SMC32 to display that many nodes: even divided by a million (to
give "meganodes") it's 10 digits. I don't think it would crash... all the
internal counters are 64-bit, which gives an upper limit of 18,446
quadrillion, still well out of reach for a single box.

Mind you, a big distributed SETI-style search with tens of thousands of
computers could top that out. A factor to bear in mind when designing the
next generation of multi-processor search engines?

MBD